Questions about Japanese spaCy models

[CodeRabbit957]

I've been playing around with the sm model (POS tagging in particular) and have a few questions...

Is there currently a (built-in) way to identify na and i adjectives?
What are all the tags included under morph for this language model? I've only come across inflection, reading and polarity so far.
What are the main limitations for this model at the moment (what it does poorly)? Are there any plans for improvements?

I've also come across some strange POS tagging for exclamation marks (!). I've seen it tagged as PUNCT, ADJ, VERB, and NUM. Why this happens, I have no idea.